Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft. Wissenschaftliche Berichte FZKA 6743

Transcription

1 Forschungszentrum Karlsruhe in der Helmholtz-Gemeinschaft Wissenschaftliche Berichte FZKA 73 INTERNATIONAL FUSION MATERIAL IRRADIATION FACILITY (IFMIF): NEUTRON SOURCE TERM SIMULATION AND NEUTRONICS ANALYSES OF THE HIGH FLUX TEST MODULE S.P. Simakov, U. Fischer, V. Heinzel, U. von Möllendorff Institut für Reaktorsicherheit Forschungszentrum Karlsruhe GmbH, Karlsruhe 00

2 Impressum der Print-Ausgabe: Als Manuskript gedruckt Für diesen Bericht behalten wir uns alle Rechte vor Forschungszentrum Karlsruhe GmbH Postfach 30, 701 Karlsruhe Mitglied der Hermann von Helmholtz-Gemeinschaft Deutscher Forschungszentren (HGF) ISSN

3 Abstract The report describes the new results of the development work performed at Forschungszentrum Karlsruhe on the neutronics of the International Fusion Materials Irradiation Facility (IFMIF). An important step forward has been done in the simulation of neutron production of the deuteron-lithium source using the Li(d,xn) reaction cross sections from evaluated data files. The developed Monte Carlo routine and d-li reaction data newly evaluated at INPE Obninsk have been verified against available experimental data on the differential neutron yield from deuteron-bombarded thick lithium targets. With the modified neutron source three-dimensional distributions of neutron and photon fluxes, displacement and gas production rates and nuclear heating inside the high flux test module (HFTM) were calculated. In order to estimate the uncertainty resulting from the evaluated data, two independent libraries, recently released by INPE and LANL, have been used in the transport calculations. The proposal to use a reflector around the HFTM to get higher dpa rate and lower spatial gradients was thoroughly investigated. The IFMIF neutronics parameters at different deployment stages regarding variation of beam intensity, footprint and energy have been estimated. Neutronik des Lithiumtargets und des Hochfluss-Testmoduls der International Fusion Materials Irradiation Facility (IFMIF) Der Bericht beschreibt neue Ergebnisse der im Forschungszentrum Karlsruhe durchgeführten Entwicklungsarbeiten für die Neutronik der International Fusion Materials Irradiation Facility (IFMIF). Ein wichtiger Schritt vorwärts in der Simulation der Neutronenerzeugung der d-li- Quelle ist durch die Verwendung von Li(d,xn)-Reaktionsquerschnitten aus ausgewerteten Kerndatensätzen erfolgt. Die dazu entwickelte Monte-Carlo-Routine und bei INPE Obninsk neu ausgewertete d-li-reaktionsdaten wurden an vorhandenen experimentellen Daten der differentiellen Neutronenausbeute aus deuteronenbestrahlten dicken Lithiumtargets geprüft. Mit der modifizierten Neutronenquelle wurden dreidimensionale Verteilungen des Neutronenund Photonenflusses, der Verlagerungs- und Gaserzeugungsraten und der nuklearen Erwärmung im Hochfluss-Testmodul (HFTM) berechnet. Um die von den ausgewerteten Kerndaten stammende Unsicherheit abzuschätzen, wurden in den Transportrechnungen zwei verschiedene kürzlich von INPE und LANL herausgegebene Bibliotheken benutzt. Der Vorschlag, den HFTM mit einem Reflektor zu umgeben, um höhere Verlagerungsraten und kleinere räumliche Gradienten zu erreichen, wurde gründlich untersucht. Die IFMIF-Neutronik- Parameter für verschiedene Ausbaustufen hinsichtlich Strahlintensität, Strahlfleckgröße und Deuteronenenergie wurden abgeschätzt.

4 Contents 1. Introduction. IFMIF reference design and computational model IFMIF reference design IFMIF/HFTM computational model d-li neutron/photon source modelling and verification Basic physical processes Approach based on M C DeLi code Approach based on M C DeLicious code Approach based on MCNPX code Comparison of code outputs with experimental data.... Evaluated nuclear data for neutron and photon transport Libraries and processing procedure INPE and LANL libraries intercomparison Results of neutronics calculations for IFMIF Lithium target Lithium target back plate High Flux Test Module HFTM volume averaged neutronics parameters Three-dimensional analyses and HFTM design optimisation Reflector effects on dpa rate and gradient Neutronics evaluation of deployment scenarios Summary and Conclusions References

5 1. Introduction The International Fusion Materials Irradiation Facility (IFMIF) is an accelerator-based D-Li neutron source with the capability to produce neutrons at sufficient energy, intensity and irradiation volume to test samples of candidate materials up to a full lifetime of anticipated use in fusion energy reactors. IFMIF should be built and operated prior or in parallel to ITER to get experimental data and theoretical knowledge in the field of irradiation effects on fusion materials necessary for DEMO and power fusion reactors projecting [1]. IFMIF is designed under coordination of the International Energy Agency (IAE) by an international team with contributions from the European Union, Japan, Russia and USA. The development of the project has gone through the Conceptual Design Activity and Evaluation in [] and is now in the next phase: Key Elements Technology [3]. At this stage, accurate assessment of the d-li neutron source parameters from basic nuclear processes is required to predict the radiation induced effects in the tested materials during the phase of detailed engineering design, optimisation and cost reduction considerations. In addition to a precise description of the d-li source term, correct simulation of the neutron transport, displacement damage, gas production and nuclear heating in different materials is necessary. In earlier work at Forschungszentrum Karlsruhe, computational tools describing neutron production in the deuteron-lithium source were developed and the generation and processing of evaluated nuclear data sets of the major structural materials for neutrons with incident energy up to 50 MeV was started [, 5]. The approach used for the source neutron simulation was based on simplified, semi-empirical models for the Li(d,xn) reaction cross section. The transport of the neutrons and the nuclear responses were calculated using the first release of the evaluated neutron data library developed in collaboration with INPE Obninsk. Limitations of those approaches and the necessity for the further improvement were expressed by their authors. Recently evaluated data libraries for complete and detailed description of the deuteron interactions with lithium up to a deuteron energy of 50 MeV have been produced in collaboration with INPE. This opens the opportunity for an advanced modelling of the IFMIF neutron source. Independently, the intense evaluation work performed at Los Alamos National Laboratory (LANL) has resulted in first releases of evaluated neutron transport data libraries for energies up to 150 MeV. This progress has motivated us to implement these achievements in the IFMIF computational analyses by a more advanced modelling the IFMIF neutron source term and performing comprehensive three dimensional neutronics calculations for the high flux test region. Some results were reported elsewhere [-10].. IFMIF reference design and computational model.1 IFMIF reference design This chapter presents an outline of the technical IFMIF features most essential for the present neutronics calculations, namely, the design and specification for the lithium target assembly and the high flux test module (HFTM). These parameters are taken from the reference conceptual design []. 1

6 IFMIF at full performance will employ two continuous-wave linear accelerators each generating 15 ma of 0 MeV deuterons with a beam footprint of 5 0 cm striking a flowing liquid lithium target, Fig. 1. A jet of lithium with 0.51 g/cm 3 density is supposed to be.5 cm thick (sufficient to stop 0 MeV deuterons) and cm wide. It is open to the accelerator vacuum and backed by a replaceable backwall 1.8 mm thick, most probably made of steel (Eurofer). Neutrons produced through the d-li interaction will irradiate materials located behind the back wall in the beam down-stream direction. In accordance with the level of neutron intensity the space will be subdivided into three regions: High, Medium and Low Flux Test Modules (Fig. 1). The High Flux Test Module (HFTM) placed directly behind the Lithium target is foreseen to be a rectangular volume of 500 cm 3, in which a displacement damage rate of more than 0 dpa (in iron) per full power year will be reachable. The Medium and Low Flux Modules are installed behind the HFTM to utilise the available flux volume for irradiation tests of materials exposed to lower neutron loadings. The highest irradiation levels in the HFTM correspond to the neutron loading of 1 5 MW/cm, which materials of the first wall of an ITER or DEMO reactor should withstand. At present, ferritic/martensitic steel (Eurofer), vanadium alloys and possibly silicon carbide ceramics are considered probable candidates. Swelling and creep phenomena, radiation hardening and embrittlement, fracture toughness, crack-growth and many other material properties are the subject of irradiation research. As is well known, the radiation effects strongly depend on the neutron dose, neutron rate, volumetric production of gas atoms like hydrogen or helium, temperature variation and spatial gradients. Thus a special investigation into the nuclear optimization of the HFTM should be undertaken with the main objective to reach the desirable irradiation level and to minimise the spatial gradient of the important nuclear responses such as damage rate. Material conversion caused by the nuclear transmutations should be kept on a negligible level. In the present reference design the HFTM container houses 7 rigs which hold the material specimens enclosed in capsules. Helium flows through the container and removes the heat induced by neutron reactions, activation and γ-radiation. The resulting power density profiles have to be counterbalanced by electric heaters, so that all specimens can be kept at defined temperatures with rather small spatial gradients. A further heater system has to control the temperature during start-up and shut-down operation of the accelerator.. Computational model The computational model used for neutronics analyses should take into account the main and essential features, i.e., geometry configuration and material specifications. The model used in the present analyses is shown in Figs. and 3. Simplifications made with respect to the original design have minor effects and can be taken into account in future analyses. The MCNP code [11, 1], which is the base of our developed computational tools, offers a variety of means for flexible and complete simulation of the real facility. Each of the two deuteron beams impinges on the target with 10 o declination angle in vertical direction. The beam spatial distribution on the target surface is shown in Fig.. It represents the beam profile, which will be formed by the accelerator optical system and delivered to the target. The related simulation routine [5] is rather flexible and is capable to model different

7 configurations. The profile shown in Fig. is a superposition of a rectangular and several Gaussian distributions, which finally result in a symmetrical Gaussian-edged profile with 0 5 cm beam footprint at half maximum level. The lithium target and its back plate were simulated as rectangular boxes.5 0 cm 3 and cm 3 filled with lithium at 0.51 g/cm 3 and Eurofer at 7.8 g/cm 3 density, respectively. Estimations performed have shown that other parts of the lithium jet or back wall, located 5-10 cm above or below the beam foot print, do not affect the irradiation level in the HFTM (i.e., rescattering of source neutrons is negligible). Since the nuclear and thermal-hydraulic module design is being performed in an iterative process, the starting nuclear calculations therefore assume a simplified HFTM configuration as shown in Fig.. The geometrical model consists of a rectangular block cm 3 filled with Eurofer with a mass density of. g /cm 3, which is 80% of the normal density to account for the space occupied by cooling gas. The advanced geometry specification options offered by MCNPC [1] make it possible to get calculation results in different form for any cell under interest: (i) averaged over its whole volume, (ii) 3-d spatial distribution inside and/or outside the cell. For the latter choice we used a lattice structure for the tallies [1]. For example, for the calculation of the 3-d spatial distribution inside the HFTM, its total volume was divided in 1000 elementary cubic segments cm 3, for every of which the neutron and photon fluxes, nuclear heating, dpa, helium and hydrogen gas production rate were calculated during one run. 3. d-li neutron/photon source modelling and verification 3.1 Basic physical processes The accurate assessment of induced radiation effects in IFMIF test modules will be affected to a large extent by the prediction quality of the spectral and angular distribution of the neutron and photon yields from the d-li source. To succeed in this the following basic physical processes should be taken into account (Fig. 3): - the energy, direction and intensity spatial distribution of the incident deuteron beam; - the slowing down of deuteron in the lithium target; - deuteron-lithium nuclear interaction resulting to emission of the neutrons and γ-rays. The profile and parameters of the dual deuteron beams come from the IFMIF design and can be simulated with the necessary accuracy. The deuteron slowing down in the lithium is traditionally accounted for by the theory of ion interaction with atoms. The most crucial and uncertain issue is the modelling of d-li reaction cross sections. Basically two competing reaction mechanisms should be considered: deuteron stripping and deuteron absorption followed by the formation of a compound nucleus with subsequent particle emission. Below, three approaches based on different Monte Carlo codes are considered and verified against available experimental data: M C DeLi, developed earlier specially for IFMIF neutronics calculations [5]; M C DeLicious, developed in the present work as a modification of M C DeLi; 3

8 - and the general transport code MCNPX [13], the latest version.1.5 of which recently became available. 3. Approach based on M C DeLi code To enable neutronics calculation for IFMIF, the Monte Carlo code M C DeLi was previously developed [5]. It simulates the configuration of dual beams incident on the lithium target surface, the direction of each beam and spatial intensity/current distribution to meet the requirements of the IFMIF project. The process of deuteron slowing down in the lithium medium is treated by the well known empirical model of Ziegler et al. [1]. The neutron production via Li(d,xn) reactions was modelled taking into account two competing reaction mechanisms: deuteron stripping and deuteron absorption by compound nucleus with subsequent emission of one or two neutrons. Free parameters of these models have been adjusted by fits to the angular-energy distributions of the neutron yields from thick lithium targets measured at 3 and 0 MeV incident deuteron energies. The subroutines simulating the deuteron beam configuration, slowing down in the lithium and Li(d,xn) reaction cross sections were linked to the MCNPB [11] code to enable neutron transport calculations for the IFMIF test modules. More recently, the M C DeLi code has been intensely tested against available experimental data for the full deuteron range from 5 to 50 MeV [10]. This test has shown that M C DeLi fails to reproduce the data for deuteron energies below 30 MeV (break-down at energies below 15 MeV) and high energy part of neutron spectrum above 0 MeV. In the deuteron energy range 3 0 MeV, for which M C DeLi was especially developed, the simplified Li(d,xn) reaction cross sections modelling offers the possibility for further improvement of the quantitative agreement with experimental data. 3.3 Approach based on M C DeLicious code The drawbacks of M C DeLi approach can be overcome if the simplified semi-empirical model for the Li(d,xn) reaction is replaced by a complete and detailed description of the deuteron interaction with lithium nuclei. This can be accomplished by the use of tabulated data from evaluated data files. Recently a newly evaluated d-li data library [15] has been developed in the collaboration of INPE (Obninsk) and FZK (Karlsruhe). Diffraction theory and a modified cascade evaporation model have been applied for calculating the cross sections and angle-energy distributions of reaction products for deuteron energies up to 50 MeV. It turns out that over this energy range, a variety of reaction channels are open for deuterons interacting with lithium nuclei. For example, as Table 1 shows, even for deuteron energies below MeV there are reactions producing neutrons on each isotope. In the INPE evaluation, all channels resulting in the emission of a specific kind of particles are summarized in cumulative particle yields, i.e. there are available in the file production cross sections such as (d,xn), (d,xγ), (d,xα) etc. Several of these cross sections for both of the lithium isotopes are shown in Fig. 5 as a function of the deuteron energy. It is clearly visible that besides elastic deuteron scattering the emission of neutrons, protons, tritium, alpha particles and γ-rays has significant probabilities. For IFMIF neutronics calculations the emission of neutrons and γ-rays is of primary importance since the other particles are short ranged and will be stopped in the lithium target.

9 Table 1. Reaction channels open in the deuteron lithium interaction up to 5- MeV incident energy, their Q values and thresholds. Neutron producing channels are emphasized by bold font. d + 7 Li d + Li Reaction Products Q, MeV Threshold, MeV Reaction Products Q, MeV γ + 9 Be α.37 0 n + α γ + 8 Be.81 0 n + 8 Be p + 7 Li α + 5 He n + 7 Be d + 7 Li 0 0 p +t + α p + 8 Li n + 3 He + α Threshold, MeV t + Li He + 5 He n +p + 7 Li t + 5 Li d +t + α d + Li 0 0 n + 7 Be d + α He + He n +p + Li n + p + t + α n + p + d + α To take full advantage of the tabulated d +,7 Li data, a special routine was developed to read and process the cross sections from the files. As a preliminary step, the ENDF/B-VI formatted data were processed by the NJOY-99 code system [1] into the ACE format suitable for the Monte Carlo sampling scheme. The sampling procedure is as follows. The deuteron track length is sampled, taking into account the total d-li interaction cross sections read from the file; then for the given deuteron energy the interaction probability with either of the Li or 7 Li nuclides is calculated according to their macroscopic cross sections; eventually the energy and angle of the generated neutron and photon are sampled. This information together with the coordinates of the d-li collision inside the lithium target cell is further used for the transport calculations of the neutrons and photons. To enable full neutronics calculations for the IFMIF neutron source, this procedure has been integrated to the M C DeLi code by replacing its d-li reaction cross section and deuteron slowing down modules. The full package was then compiled with the MCNPC routines [1], resulting in the advanced M C DeLicious code which is capable of using of evaluated data files for the neutron and photon generation through the d-li reaction. The correctness of the NJOY processing and Monte Carlo sampling by the M C DeLicious code was checked by the following procedure. A rather thin target of thickness 0.1 mm filled with the 7 Li isotope was considered. The incident deuteron energy was selected in such a way that the energy of the collided deuterons does not deviate very much from an energy point given on the file (e.g., E = 0.00 ± 0.05 MeV). For this energy, the data can be derived from the original INPE file in ENDF format without processing. The neutron and γ-ray yields calculated by M C DeLicious code were converted to double differential cross sections. The comparisons of results for 7 Li(d,xn) and 7 Li(d,xγ) reactions are shown in Fig. and 7 at 0 MeV deuteron energy. Good agreement is found between the data stored on the original INPE file and those calculated by the M C DeLicious Monte Carlo code using the data processed by NJOY. The remaining inconsistencies are supposed to result from the 5

10 differences of the energy and angular bins, interpolation during NJOY-code processing and sampling. It is worthwhile to note that an evaluation of neutron induced cross sections for the interaction with lithium nuclides was provided by INPE too [15]. Using these evaluated data, it is possible to assess the neutron multiple scattering effect in the lithium target itself as well as the yield of secondary γ-rays due to Li(n,xγ) reactions. 3. Approach based on MCNPX code The Monte Carlo code MCNP (versions a to c) was developed to calculate the transport of neutrons, γ-rays and electrons as a source radiation. The next major extension of this code, MCNPX, enables to track all (including charged) particles up to energies of a few GeV. The presently available version.1.5 [13] is not yet capable of using cross sections from evaluated data libraries for charged particles. Instead, it uses various nuclear models, which originally were developed for the intermediate and high energy physics domain. For relatively low incident energies (< 150 MeV) and light target nuclei MCNPX assumes, as default and the best option, the ISABEL intranuclear cascade model. From the methodological point of view MCNPX.1.5 resembles the M C DeLi code: both use built-in analytical models for the d + Li interaction. It is questionable, however, if the ISABEL model is able to predict the d-li neutron source characteristics with sufficient accuracy compared to M C DeLi, which was developed and fitted for this particular reaction and energy range. Nevertheless it is certainly reasonable to start testing MCNPX, since future extensions of the code are announced to be capable of using evaluated charged particle data libraries. 3.5 Comparison with experimental data There have been a few independent measurements of neutron yield spectra from deuteron bombardment of thick lithium targets [17-5]. The list of experiments in chronological order and the parameters most important for the present analyses are shown in Table (for a comprehensive review of the experiments, their intercomparison and more details see [10] and related original publications). To simulate these experiments we used a model adapted to a typical experimental configuration: a parallel monoenergetic beam of 1 cm in diameter, a lithium target of cm in diameter and cm thickness, and a neutron detector recording the flux at 500 cm distance from the target. The problem of multiple neutron scattering on the target material (lithium) was investigated by calculations with the newly developed M C DeLicious code and n-li data from INPE evaluations. In Fig. 8 two neutron spectra are compared: one calculated with the target cell filled by Li at normal density and the other with a void target cell (generation of the source neutrons is the same in both cases). It is seen that the neutron-lithium collisions result in neutron absorption in the target around 0% for neutron energies below 0 MeV and reaching the maximum value of 70% in the vicinity of the 7 kev resonance in the n-li total cross section. The spectrum averaged correction for neutron absorption in the target material amounts to 10-15% of the neutron flux and should be therefore taken into account when comparing with experimental results. Fig. 9 compares the experimental total and forward neutron yields with the M C DeLi, M C DeLicious and MCNPX calculations as a function of deuteron energy. Note that the experimental data were measured using different energy thresholds ranging from 0.3 to

11 3.5 MeV (Table ). Therefore, the related calculations were performed with a neutron energy threshold of MeV. One can conclude that M C DeLicious with evaluated d-li data from INPE predicts the energy dependent neutron yield better than M C DeLi or MCNPX do. It is also seen that M C DeLi is not capable of reproducing the experimental data below 0 MeV, whereas MCNPX underestimates the neutron yield by a factor of. Table. Parameters of the measurements of neutron spectral and angular distributions from thick Li targets bombarded with deuterons. No First Author Year of Publ. Laboratory, Country Target E d, MeV Θ, degrees E thr, MeV Ref. 1 V.K. Daruga 198 Inst. of Physics and Power Engi- Li 0 o 1.8 [17] A.N. Weaver A.N. Goland 1975 H.I. Amols C.E. Nelson 1977 M.A. Lone M.J. Saltmarsh D.L. Johnson M. Sugimoto 1995 neering, Russia Lawrence Livermore Laboratory, USA Naval Research Laboratory, USA Fermi National Accelerator Laboratory, USA Triangle University Nuclear Laboratory, USA Chalk River Nuclear Laboratory, Canada Oak Ridge National Laboratory, USA University of California, USA Japan Energy Research Institute, Japan Li 5, 9, 1, 1, 19 Li 13.,19.0,.8,8.9, o, 10 o, 18 o,.5 5 o, 3 o 1. [18] 0 o, 5 o, 10 o, 15 o, 3 [19] 0 o Li 35 0 o 5 [0] 7 Li 8, 1, 15 0 o, 10 o, 0 o, 30 o, 5 o 1 [1] 7 Li 1.8, 18, 3 0 o, 10 o, 0 o, 30 o, 0.3 [] 0 o Li 0 0 o, 7 o, 15 o, 30 o, 5 o, 0 o, 90 o [3] Li 35 0 o, o, 8 o, 1 o, 0 o, 30 o, 5 o, 70 o, 105 o, 150 o Li 3 0 o,5 o,10 o,15 o,0 o, 30 o,0 o,50 o,0 o, 70 o,80 o,90 o,100 o, 110 o,10 o,130 o, 10 o, 150 o 1 [] 1 [5] The average energy of neutrons escaping from the thick lithium target in forward direction versus the deuteron energy is depicted in Fig. 10. It is seen that M C DeLi and M C DeLicious exhibit results close to each other and to the experimental data for deuteron energies above 15 MeV, whereas MCNPX overestimates the average neutron energy by 10-50% for all deuteron energies under consideration. The angular distributions of the emitted neutrons are shown in Fig. 11 for incident deuteron energies 15 to 0 MeV. For deuteron energies 35 and 0 MeV, the three approaches considered in this work give results fluctuating near the experimental data. The MCNPX results, however, reveal a somewhat weaker anisotropy in the forward hemisphere, which is 7

12 important for the IFMIF application. With decreasing incident deuteron energy the code outputs deviate more and more from each other and also disagree with the experimental results. At MeV deuteron energy, only M C DeLicious provides fairly good agreement with the measured data. Double differential data spectral neutron yields at different emission angles are presented in Fig. 1 and 13 for the two deuteron energies 3 and 0 MeV respectively. These figures again show that M C DeLicious agrees better than M C DeLi and, in particular, MCNPX with the experimental results. It is also revealed that a further essential improvement was achieved with the INPE data in the high energy part of the spectra (E n > 30 0 MeV). These neutrons, produced in the exothermic 7 Li(d,n) 8 Be reaction with Q = 15 MeV (Table 1), were not taken into account in the M C DeLi approach. The yield of these high energy neutrons is relatively small ( 0.8%), but they will have more significant importance for the activation and shielding behaviour. The principally new possibility, which the updated code M C DeLicious has opened, is the assessment of the γ-ray yield emitted from the lithium target. There are two physical processes which produce photons in the lithium: first, the primary reaction Li(d,xγ) and, second, neutron induced secondary reactions such as the inelastic scattering of source neutrons on lithium nuclei, i.e., the Li(n,xγ) reaction. Both of these productions paths were analysed by means of M C DeLicious calculations using the evaluated INPE data for d +,7 Li and n +,7 Li interactions. Total and forward (angular distribution is isotropic) photon yields from a thick lithium target bombarded by deuterons, as predicted by M C DeLicious, are shown in Fig. 9 versus the incident deuteron energy. Since experimental data on the Li(d,xγ) reaction are not available, these theoretical predictions cannot be verified at present. The comparison with the neutron yield reveals that the photon yield is one to two orders of magnitude lower. The energy distribution of primary γ-rays, as shown in Fig. 1, consists of two parts: low energy photons (E γ < 5 MeV), resulting from Li(d,d γ), (d,n γ), (d,p γ) reactions; and high energy photons from deuteron capture, i.e., Li(d,γ) reaction. As Fig. 15 shows, the high energy photons amount to 10 - of the total yield and have an average energy 0 MeV, much larger than the energy ( 0.8 MeV) of dominating low energy photons. Although the fraction of high energy photons is relatively small, they will result in activation and transmutation reactions in the IFMIF test modules, since the photo-nuclear reactions have thresholds of a few MeV. The contribution of the neutron induced γ-ray production depends on the size and mass of the lithium target. For the Li target of the size mentioned above (cylinder cm, Li-density 0.51 g/cm 3 ), it amounts to about 8% of the deuteron induced γ-ray yield. The photon energy spectrum, depicted in Fig. 1, reveals that most of the γ-rays have energies below 5 MeV (the discrete lines come from the de-excitation of excited states of the residual nuclei). Nevertheless, also γ-rays with very high energies (10-50 MeV) are produced by neutron capture reactions on target nuclei. Based on comparisons with available experimental data for deuteron bombarded thick lithium targets, we come to the following conclusions. The updated M C DeLicious code, which is now able to use the INPE evaluation of d-li cross sections, predicts the neutron source characteristics better than M C DeLi does. M C DeLicious is also able to calculate the photon production in the lithium target, but this could not be verified due to the lack of related 8

13 experimental data. The charged particle transport code MCNPX (version.1.5) underestimates the neutron yield typically by a factor of and fails to calculate the γ-rays yield.. Evaluated nuclear data for neutron and photon transport.1. Libraries and processing procedure Evaluated data for neutron transport calculations have been intensely developed and tested worldwide for many nuclei up to the neutron energy of 0 MeV. As a result such libraries as ENDF, EFF, FENDL, JENDL, BROND and others are now used in many applications. To perform the IFMIF neutronics analyses these libraries need to be extended above 0 MeV, since the d-li source at 0 MeV deuteron energy produces neutrons with energies up to 55 MeV. Besides the differential cross sections needed for transport calculations the evaluated data libraries should provide as well engineering responses, i.e., nuclear heating, gas production and displacement damage rates. At the moment the following data evaluations that meets these requirements are available: INPE 50 developed at the Obninsk Institute of Nuclear Power and Engineering in collaboration with FZK for the nuclides: 1 H,,7 Li, 1 C, 3 Na, 8 Si, 39 K, 51 V, 5 Cr, 5 Fe and neutron energies up to 50 MeV [15, 7, 8]; LANL 150 developed at Los Alamos National Laboratory for the nuclides: 1 H, 1 C, 1 O, 7 Al, 8,9,30 Si, 31 P, 0 Ca, 50,5,53,5 Cr, 5,5,57,58 Fe, 58,0,1,, Ni, 3,5 Cu, 93 Nb, 18,183,18,18 W, 0,07,08 Pb and neutron energies up to 150 MeV [9]. To make them available for the Monte Carlo calculations, the original data files in ENDF format must be processed to the ACE format. While the LANL library is already distributed in this format, the processing of the INPE library has been performed by the authors with the NJOY-99 code system [1]. During the processing additional kinds of data were calculated and allocated in the files. The most important one is the heating due to nuclear reactions, which can be conveniently divided into neutron and photon heating. The neutron heating arises from the kinetic energy of charged products (KERMA factors) and is deposited locally at the point of neutron induced reaction. The HEATR module of NJOY generates heat production cross sections and adds them to the ACE file. The photon heating is proportional to the flux of secondary photons transported from other sites of previous neutron interactions. It is estimated during the coupled neutron-photon transport calculations by MCNP (for this purpose the photon atom interaction cross section evaluated data from ENDF library [30] are used in the energy range 1 kev 100 MeV). The other important design consideration is the damage caused in materials by neutron and photon irradiation. The total energy available to cause displacements is calculated by NJOY from the total kinetic energy of recoil nuclei reduced by the energy lost for electronic excitations (accounted for by the so called Robinson partition function [31]). Eventually, the displacement damage cross sections depends on the damage energy production cross section σ Edp and the energy E d required to displace an atom from its lattice position: σ dpa = 0.8 σ Edpa / E d 9

14 The values for the effective displacement threshold E d are chosen to represent experimental results or derived from theoretical calculations. Recommended values can be found in the literature [3]. For some elements, playing key role in the present calculations, the displacement thresholds are presented in Table 3. During the transport calculation the damage rate for the particular material in the selected geometrical cell is computed by folding the estimated spectral neutron flux with the displacement damage cross section. Table 3. Typical values for the atomic displacement energy E d needed to compute damage cross sections. Element E d, ev Element E d, ev Be 31 Co 0 C 31 Ni 0 Mg 5 Cu 0 Al 7 Zr 0 Si 5 Nb 0 Ca 0 Mo 0 Ti 0 Ag 0 V 0 Ta 90 Cr 0 W 90 Mn 0 Au 90 Fe 0 Pb 5.. INPE and LANL libraries intercomparison As an example of NJOY processed evaluated data, Fig. 1 shows cross-sections from two libraries, LANL and INPE, for one of the most important nuclides, 5 Fe. It is seen that these two evaluations reasonably agree with each other, increasing our confidence in the neutronics calculations. The most significant disagreement manifests itself for the neutron energies 0 to 50 MeV, where the neutron yield from Li(d,n) source, as shown in Fig. 1, rapidly decreases. Note also that the INPE evaluated cross sections do not demonstrate resonance behaviour in the energy range 0. MeV, as LANL does. As could be expected, this does not play a significant role in the present calculations because of the relatively small thickness (compare with neutron mean path) of the materials under consideration. Both libraries show that the damage cross sections, gas and heat productions caused by neutrons are increasing function of the energy, underlining the importance of accurate prediction of the high energy part of the source spectrum. Fig. 17 compares the neutron and photon cross sections and heat deposition in 5 Fe. The heating due to photons dominates over neutrons at least by a factor of 10 and is sensitive to both high and low energy photons. On the other hand, the typical photon flux in the HFTM, produced in the lithium target and neutron inelastic collisions, is a few orders of magnitude less than the neutron flux. Thus the final contribution to the total heating will be the balance of these two factors. The cross sections for some other nuclei, which constitute the materials planned to be irradiated in the IFMIF, are shown in Fig It is interesting to note that the proton production rate exceeds the helium production for every nuclide except carbon, for which the (n,α) reaction is favourable. Such behaviour in the case of carbon obviously is the result of the cluster-like structure of the 1 C nucleus and its lower threshold for alpha production (. MeV) as compared to proton production (13. MeV). Due to this property a higher rate of 10

15 helium production may be expected in carbon containing materials. For sodium and potassium, which are potential coolant materials for the test module, the evaluated data in the high energy range are incomplete (Fig. and 3). The only available evaluation is that performed at INPE, but it does not contain the total photon, proton and alpha-particle production cross sections. To demonstrate their behaviour up to 0 MeV neutron energy the evaluated cross sections from ENDF/B-VI library are depicted. Besides neutrons, γ-rays can also produce nuclear transmutations and cause damage in materials due to photo-nuclear reactions. The related cross sections and energy spectra of reaction products were only recently tabulated in specific photo-nuclear evaluated files, e.g., the IAEA or LANL Photonuclear Data Libraries [33, 3], which are complete with respect to radiation transport calculations. Fig. shows some γ-ray induced nuclear reactions cross sections for 5 Fe from the IAEA library. Note that photo-nuclear reactions due to their high energy threshold are possible only with hard photons. The fraction of such photons is relatively small. This is why the damage and gas productions due to photons are expected to be negligible. Until now no transport code able to use photonuclear data in a fully-coupled manner is available. Taking this into account we have estimated the contribution of photonuclear reactions only for the gas production by folding the γ-ray spectral flux averaged over the cell of interest with the (γ,xp) and (γ,xα) cross sections. 5. Results of neutronics calculations for IFMIF 5.1. Lithium target Results of neutronics calculations for the lithium target obtained with the M C DeLicious code are listed in Table. Given the energy of each beam is 0 MeV and the current 15 ma, the total power delivered to the lithium jet is 10 MW. This beam will produce neutrons per second (in π solid angle), thus 131 kw will be carried out from the target as neutron kinetic energy. Every 100 deuterons will produce 7. neutrons (with mean energy 7.3 MeV) and 1. photons (mean energy 0.8 MeV) due to Li(d,xn) and Li(d,xγ) reactions. The number of secondary photons, produced in neutron inelastic collisions with the lithium jet, amounts to 0.1 per 100 deuterons, i.e. one order of magnitude less than the primary γ-ray source. The deuteron penetration range into the lithium and the energy deposition depend on the rate of energy loss due to deuteron interaction with atoms. For the estimation of the energy loss profile we used the MCNPX code [13], which is able to simulate energy and angle changes resulting from both nuclear and electronic interactions. Given the lithium density of 0.51 g/cm 3, the dependence of the deuteron energy on the penetration depth and the energy deposition profile (at the total beam current of 50 ma) along the beam symmetry line were calculated. The results depicted in Fig. 5 show that the maximum of the beam energy is deposited near the end of the deuteron track, reaching there the extremely high value of more than 300 kw/cm 3. To prevent local lithium boiling, the accelerator beam transport system is planned to introduce an energy dispersion of the incident deuterons. A Gaussian energy distribution with a root mean square (rms) parameter of 0.5 MeV, as shown in Fig. 5, efficiently smears the energy deposition in the vicinity of the peak, resulting in a reduction of the maximum by more than a factor of. It is worthwhile to note that the energy dispersion of the incident deuterons slightly increases (by less than 10%) the beam penetration range in the target. 11

16 Table. Deuteron beam and d-li source parameters, nuclear induced effects in the lithium target back plate and HFTM. Parameter Value Li + d Nuclear Radiation Source Accelerated Deuteron Energy 0 MeV Beam Current beams 15 ma = 50 ma Beam Footprint 0 5 cm Beam Power 10 MW Li Target thickness.5 cm Li Target density 0.51 g/cm 3 Mean Deuteron Energy in Li-target 5. MeV Deuteron Induced Nuclear Reactions Li(d,xn) Li(d,xγ) Number of n or γ per 1 d 0.07 n/d 0.01 γ/d Source Intensity (π) n/s γ/s Source Radiation Power (π) 131 kw. kw Mean n- or γ- Energy (π) 7.3 MeV 0.8 MeV Lithium Target Back Plate (BP) Size cm 3 Front Surface 100 cm Volume 18 cm 3 Material Eurofer (Fe-88.9%, Cr 9.%, C-.9%, ) Material Density 7.8 g/cm 3 Average Neutron Flux n/cm /s Average Neutron Current n/cm /s Average Neutron Energy 8.1 MeV Neutron and Photon total flux Power kw Neutron & Photon Energy Loadings Density MW/m Average Gamma-ray Flux γ/cm /s Average dpa-rate 5 dpa/fpy Average Hydrogen Production appm/fpy Average Helium Production 5 appm/fpy Total Heat Production 1 W Average Heat Production Density 3 +. W/cm 3 High Flux Test Module (HFTM) Size cm 3 Volume 500 cm 3 Material Eurofer (Fe-88.9%, Cr 9.%, C-.9%, ) Material Density. g/cm 3 (80% of normal) Average Neutron Flux n/cm /s Average Gamma-ray Flux γ/cm /s Neutron Energy Load. ( ) MW/m Average dpa-rate 9 dpa/fpy Average Hydrogen Production appm/fpy Average Helium Production appm/fpy Total Heat Production 7.0 kw Average Heat Production Density W/cm 3 Comments: in right-hand column the first additive in the sum stands for the contribution from the Li(d,xn) reaction, the second for Li(d,xγ). 1

17 5.. Lithium target back plate The lithium target back plate obviously must withstand the highest irradiation loadings in the facility. As mentioned above, a stainless steel (Eurofer) is regarded as the most probable material for the target back wall. For the estimation of its replacement period we calculated the main neutronics and irradiation parameters with the M C DeLicious code. Table lists the results averaged over the most strongly irradiated volume cm 3 of the back plate. It shows that the back wall material should withstand the neutron flux n/cm /s and the energy load density 9 MW/m, which is equivalent to 50 dpa per full power year. The neutron irradiation will heat the back wall with W/cm 3 power density. The spatial distributions of the main nuclear responses are shown in Fig. as contour maps in the X-Y plane. The main topological features reflect the beam density profile in horizontal and vertical directions at the lithium target surface. A relatively large region of moderately flat distribution with highest density of nuclear induced radiation effects is clearly seen in the centre part of the back plate, and regions of high gradients at its edges. The largest gradients of 50%/cm are observed in the horizontal direction at the beam footprint edge near x = 10 cm. This could result in the considerable tensions and degradation of mechanical properties in the back wall material there. 5.3 High Flux Test Module HFTM volume averaged neutronics parameters The characteristics of the neutron and photon fluxes, calculated with the M C DeLicious code and averaged over the HFTM volume, are listed in Table. The neutron flux, which is the sum of primary and scattered neutrons, has a value of n/cm /s. The γ ray flux is approximately two times less, 90% of which are photons born in the neutron interaction with the HFTM materials. Table also shows results for radiation induced effects in Eurofer: atom displacement, gas generation and heat production. In particular, it is seen that the assessed averaged displacement rate (9 dpa/fpy) meets the IFMIF reference design requirement to have a 0.5 liter volume with at least 0 dpa/fpy. The average heat deposition in the HFTM is about 1 W/cm 3, comparable with the nuclear heating power density in the first wall of fusion reactors. The total nuclear heating power at the level of 7 kw will require an efficient cooling system for the test module. The contribution of the photons to the material damage effects was estimated by folding the calculated γ-ray spectra in the HFTM with photo-nuclear reaction cross sections. As Table shows, engineering responses to photons turn out to be less than 5% for the heating and 10-3 % for the gas production in comparison with neutrons. Such low sensitivity to photons is due to the softness of the photon spectrum and high thresholds of the relevant reactions (Fig. ). Regrettably it is impossible to perform a quantitative assessment of the γ ray contribution to the atom displacement damage, because complete photon-nuclear data libraries are not available. The neutron energy spectrum averaged over the HFTM is shown in Fig. 7. It is a product of direct d-li source neutrons and neutrons scattered by the test module, lithium target and back wall. The comparison of M C DeLicious and M C DeLi results reveals some differences in different energy regions, mainly above 0 MeV, where M C DeLicious demonstrates its ability 13

18 to predict the neutron yield from the 7 Li(d,xn) reaction. The fraction of such neutrons in the total spectrum is 0.8%. The HFTM/IFMIF neutron spectra are compared as well in Fig. 7 and Table 5 with the spectral fluxes in the first wall of the ITER and DEMO fusion reactors [35]. The latter two have similar shape but different absolute values, reflecting the fact that the first wall of the DEMO (loading of 3.5 MW/m ) will be exposed to a higher neutron flux than ITER (1. MW/m ). Both spectra contain the well-defined D-T fusion neutron peak, amounting to 1% of the total flux, and a broad spectrum of inelastically scattered and slowed-down neutrons. The mean energy is around 3.5 MeV, whereas there are practically no neutrons exceeding the fusion cut-off energy of 1. MeV. In the HFTM of IFMIF an even higher level of the neutron loading (. MW/m ) will be achieved due to the higher mean energy of neutrons, 7 MeV. The spectral distribution, as Fig. 7 shows, has a broad maximum near 1 MeV and a tail up to 55 MeV. The fraction of neutrons with E > 1. MeV amounts to about 50% (Table 5); this constitutes the spectral difference between IFMIF and fusion reactors. These high energy neutrons produce 30 to 80% percent of the displacements and gas atoms. Fortunately, as Table 5 shows, the H/dpa and He/dpa ratios, which affect the radiation behaviour of materials, are comparable with the values calculated for the reactor first wall [35]. Another reason for criticism of IFMIF has been the difference in the cascades produced by primary knock-on atoms with different kinetic energies. Recent theoretical studies [3] based on molecular dynamics simulation have shown that a large cascade of vacancies and interstitials created by a high energy knock-on atom during its evolution rather soon collapses and disintegrates into smaller ones. This means that a difference in the primary knock-on atom energy results in a different number of defects but does not touch the basic mechanisms of radiation damage Three-dimensional analyses and HFTM design optimization The neutronics calculations presented above considered the HFTM as one single cell. Since this cell is located close to the neutron source, the neutron flux and induced nuclear effects will strongly depend on the position inside the test module. On the other hand they will be affected by neutrons scattered on materials which inevitably will surround the test module or could be specially allocated around it to improve some neutronics features, e.g., to minimize spatial gradients, optimise the HFTM geometrical configuration etc. Such analysis can be accomplished by calculating three dimensional distributions inside and outside the HFTM (the method used was described in section.). The spatial distributions are presented in Figs. to 33 along the depth into HFTM (Z-axis or deuteron beam direction as shown in Fig. ) starting from selected points at the front plane: geometrical centre of the HFTM, top side centre, corner and right side center. Calculations have been performed with two libraries (LANL and INPE) to test the sensitivity of the results to the evaluated data. The comparison has shown that the neutron and γ-ray fluxes and the dpa-rate agree with each other within 10%. Larger discrepancies, 10 to 0%, are found for the heating, hydrogen and helium production rates. Obviously this is the result of the differences in the KERMA factor, (n,xp) and (n,xα) reaction cross sections as seen in Fig. 1. 1

19 Table 5. HFTM/IFMIF neutron flux parameters and radiation induced effects in different materials, comparison with irradiation in the first wall of ITER and DEMO. Facility IFMIF/HFTM ITER DEMO Material Steel Steel Vanadium Ceramic Iron Iron Parameter Eurofer F8H-mod VTiCr SiC Fe Fe Wall Load, MW/m (Total) (E > 1. MeV) 3.0 ( 7 %) - - n-flux, 10 1 /cm /s (Total) (E > 1. MeV) 1.1 (19%) - Mean Neutron Energy, MeV (E > 1. MeV) 1 - dpa, 1/fpy (Total) (E > 1. MeV) 10 (37%) 10 (0%) 9.9 (35%) 5. (3%) - - Volume (dpa > 0/fpy), cm H-production, appm/fpy (Total) (E > 1. MeV) 100(70%) 950(83%) 50 (79%) 70 (0%) - - Ratio H/dpa (Total) (E > 1. MeV) He-production, appm/fpy (Total) (E > 1. MeV) 10 (3%) 90 (7%) 9 (81%) 1770(8%) - - Ratio He/dpa (Total) (E > 1. MeV) Nuclear Heating, W/cm 3 (Total) (E > 1. MeV). (%)

20 As one can see, the dpa-rate and gas production are monotonically and rapidly decreasing functions of the depth. This is explained by the dependence of these parameters on the neutron flux, which rapidly decreases with increasing distance from the neutron source. In contrast, the heat production in the HFTM has a lower gradient and is even flat in the front layer of cm. The reason is the dominating contribution of photons to the nuclear heating, the spatial distribution of which has a broad maximum along the beam direction. Fig. 3 shows that the photon flux, though it is -3 times less than the neutron flux, heats the Eurofer with -3 times as much power. This reflects the capability of photons to release much more heating energy in matter than neutrons with the same energy will release (Fig. 17): one order of magnitude for energies above 1 MeV and up to five orders at lower energies Reflector effects on the dpa rate and gradient To attain a higher average displacement rate and flatter spatial distribution in the high flux module, it was proposed to use a reflector around it, that will return some fraction of neutrons back into the HFTM []. For quantitative assessment of the reflector effect we used the computational model shown in Figs. and 3. The HFTM was surrounded from all sides (except the side towards the lithium target, the effect of which was found to be negligible) by plates filled with the material at 0-90% of normal density to take into account the heat removing channels. The following elements were considered as candidates for the reflector: aluminium, carbon, lead, nickel, iron (Eurofer), beryllium and tungsten. The radiation induced effects inside the HFTM were assessed for different reflector thickness. It was found that they increase as the reflector becomes thicker, but reach saturation at 10 cm. For this thickness the most prominent reflecting effect was demonstrated by Eurofer, tungsten and beryllium. Figs. 35 and 3 show calculation results for dpa and heating rates for the bare and surrounded test module along two lines: the HFTM symmetry axis (center) and its corner, where the reflector effects are expected to be strongest. It is clearly visible that the relative increase of displacement damage and heat power density on the periphery amounts to 0-50%, that is, times more than in the central region of the HFTM, 10-0%. This means that reflector indeed results in improved irradiation conditions in the HFTM: the dpa rate increases, and the spatial gradient decreases. As for the selection of the more optimum material for the reflector, one can see that Be, W and Eurofer equally affect the displacement damage level, while Eurofer among them has the greatest effect in the flattering the heat production distribution in the test module. This and other favourable properties of Eurofer (like the well investigated behaviour of its mechanical properties under irradiation and its reduced activation) make it one of the most reasonable candidate materials for the reflector. We have selected it for the further, more detailed investigation. With a 10 cm thick Eurofer reflector the spatial gradient along the deuteron beam direction (in which it has the maximum value) ranges from 18 to 0%/cm for the displacement rate and to 13%/cm for the nuclear heating density. The influence on other neutronics parameters is shown in Figs. 37 to 1 and Table. In particular, this table shows that neutron and photon fluxes averaged over the HFTM volume increase by 0-0% due to the reflector. The displacement rate, sensitive mainly to the high energy neutron flux, increases by 8%, and the volume available for material testing at 0 dpa/fpy by 1%. Maximum changes up to 5% 1